基于三维点云模型获取单木树高和胸径的研究

StudyonObtainingHeightandDBHofSingleTreBasedon3DPointCloudModel

摘要:

使用无人机和智能手机分别从空中和地面拍摄的样地林分影像构建三维点云模型,并从三维点云模型中获取样 地内单木树高和胸径参数 。本研究以池杉人工林为研究对象,利用 PhotoScanAgisoft软件对无人机倾斜摄影和智能手机 近景摄影的样地影像进行三维重建,通过对齐照片、控制点刺点、对齐优化、建立密集点云等步骤,构建出与样地实景相符 的三维点云模型;通过LiDAR360软件从样地三维点云模型中获取单木的树高和胸径参数,将其与实地测量获取的单木树 高和胸径参数进行对比分析 。利用无人机和智能手机影像构建的三维模型可以满足《数字航空摄影测量测图规范》的精度 要求 。通过实测数据和点云数据获取的树高和胸径的平均差值分别为-0.9 m和-0.8 cm,平均相对误差分别为 5.4%和 7.1% 。 以实测数据作为 自变量x,以点云数据作为因变量 y,树高和胸径回归模型的R2 分别为 0.8095和 0.9184。将倾 斜摄影和近景摄影的点云模型统 一在同 一 空间参考基准下可构建出与样地实景相匹配的三维点云模型,从样地三维点云 模型中获取的单木树高和胸径与实地测量结果具有较好的线性相关性,本研究所使用的方法可以代替传统林业调查手段 应用于人工林。

Abstract:

The3Dpointc1oudmode1wasconstructedfromaeria1andgroundstandimagestakenbyuavandsmartphone respective1y,andtheheightandDBH parametersofindividua1tresinthesamp1ingareawereobtainedbythe3D point c1oudmode1.Thep1antationTaxodiumchinensisvar.imbricatum wasstudiedinthisstudy,PhotoScanAgisoftsoftware wasusedtoconduct3Dreconstructionofthesamp1esiteimagesfromUAVob1iquephotographyandsmartphonec1ose- rangephotography.Throughthestepsofphotoa1ignment,contro1pointthorn,a1ignmentoptimization,andestab1ishment ofdensepointc1oud,a3Dmode1consistentwiththerea1sceneofthesamp1esitewasconstructed;TheheightandDBH parametersofindividua1treswereobtainedfromthepointc1ouddataofthe3Dmode1ofthesamp1ep1otbyLiDAR360 software,andthencomparedwiththeheightandDBH parametersobtainedbyfie1dmeasurement.The3D mode1con- structedbyuavandsmartphoneimagescanmettheacuracyrequirementsofDigita1Aeria1PhotogrammetryMapping Specification.TheaveragediferenceoftreheightandDBHobtainedfromthemeasureddataandpointc1ouddatawere -0.9 m and-0.8 cm,respective1y,andtheaveragere1ativeerorswere5.4% and7.1%,respective1y.Withmeasureddata asindependentvariab1ex andpointc1ouddataasdependentvariab1eR(Y) oftreheightandDBH regresionmode1s were0.8095and0.9184,respective1y.A3Dmode1matchingtherea1sceneofthesamp1esitecanbeconstructedbasedon thepointc1ouddataofob1iquephotographyandc1ose-rangephotographyunderthesamespatia1reference.Theheightand DBH ofindividua1tresobtainedfromthepointc1ouddataofthe3Dmode1ofthesamp1esitehaveagood1inearcore1a-tionwiththefie1dmeasurementresu1ts.Themethodusedinthisstudycanrep1acethetraditiona1forestrysurveymethod andbeapp1iedtotheartificia1forest.